Rubix CLI examples

Worked examples that combine the CLI commands into useful everyday flows.

Morning health check

Start your day with one prompt against production.

rubix chat --prompt "Summarise production health: open incidents, pending actions, deployments in the last 12 hours."

Combine with your shell startup if you want it every time you open a terminal:

# In ~/.zshrc or ~/.bashrc
alias morning-health='rubix chat --prompt "Summarise production health: open incidents, pending actions, deployments in the last 12 hours."'

On-call triage

When a page fires, drop into an interactive session scoped to the right environment.

rubix

Then:

/environments
(pick prod-eks)

What is the state of the payments namespace right now?
Which services are degraded?
What changed in the last hour?

Chat keeps the context across follow-ups. No need to repeat the namespace.

Investigate a specific resource

rubix chat --prompt "Why is pod api-gateway-7d4b9c8f6d-abc12 in prod-eks restarting?"

Single-turn, cited answer, back to your shell.

Post-deploy verification

After a deploy, compare to the last stable window.

rubix chat --prompt "Compare payments-api metrics in prod-eks for the last 10 minutes to the previous 30 minutes. Flag any metric more than 20% worse."

Or turn this into a custom skill named post-deploy-verification, then:

rubix chat --prompt "/skills post-deploy-verification payments-api"

Same result, less typing, consistent across everyone on the team.

Weekly cost review

rubix chat --prompt "For the last 7 days, which services across all production environments ran more than 15% above their 28-day baseline cost? Top 10 only."

Pipe the output if you want to paste into a doc or Linear ticket:

rubix chat --prompt "Weekly cost review..." > /tmp/cost-review.md

Send shell output into chat

You run a command, the agent reads the output.

kubectl get pods -n payments | grep -v Running

Drop into chat:

rubix

Then inside chat:

/send
Why are these pods not Running?

The agent sees both the command output and your question.

Resume an investigation later

List your recent sessions:

rubix sessions

Pick one and resume:

rubix chat --session-id "abc-def-123"

Or from inside chat: /resume and choose.

CI usage

Non-interactive checks from GitHub Actions, GitLab CI, Jenkins, or similar.

# .github/workflows/post-deploy.yml
- name: Verify deploy
  env:
    RUBIX_API_KEY: ${{ secrets.RUBIX_API_KEY }}
    RUBIX_ENV_ID: prod-eks
  run: |
    npx @rubixkube/rubix chat --prompt "/skills post-deploy-verification ${{ env.SERVICE_NAME }}"

The RUBIX_API_KEY env var skips the device-code prompt. RUBIX_ENV_ID pins the environment for that invocation.

Handoff summary

End your shift with a concise handoff message.

rubix chat --prompt "/skills on-call-handoff" > /tmp/handoff.md
cat /tmp/handoff.md

Or post directly to Slack via a webhook if you have that wired up.

Where to go next

Commands reference

Every flag and slash command.

Custom skills

Bake your best prompts into reusable skills.

Troubleshooting

If something is not working as expected.

​Morning health check

​On-call triage

​Investigate a specific resource

​Post-deploy verification

​Weekly cost review

​Send shell output into chat

​Resume an investigation later

​CI usage

​Handoff summary

​Where to go next